Copyright 1981 by the American Psychological Association, Inc. 0022-3514/81 /4103-0586S00.75
Journal of Personality and Social Psychology 1981, Vol. 41, No. 3, 586-598
Cultivating Competence, Self-Efficacy, and Intrinsic Interest Through Proximal Self-Motivation Albert Bandura and Dale H. Schunk Stanford University The present experiment tested the hypothesis that self-motivation through proximal goal setting serves as an effective mechanism for cultivating competencies, self-percepts of efficacy, and intrinsic interest. Children who exhibited gross deficits and disinterest in mathematical tasks pursued a program of self-directed learning under conditions involving either proximal subgoals, distal goals, or no goals. Results of the multifaceted assessment provide support for the superiority of proximal self-influence. Under proximal subgoals, children progressed rapidly in self-directed learning, achieved substantial mastery of mathematical operations, and developed a sense of personal efficacy and intrinsic interest in arithmetic activities that initially held little attraction for them. Distal goals had no demonstrable effects. In addition to its other benefits, goal proximity fostered veridical self-knowledge of capabilities as reflected in high congruence between judgments of mathematical self-efficacy and subsequent mathematical performance. Perceived self-efficacy was positively related to accuracy of mathematical performance and to intrinsic interest in arithmetic activities.
Much human behavior is directed and sustained over long periods, even though the external inducements for it may be few and far between. Under conditions in which external imperatives are minimal and discontinuous, people must partly serve as agents of their own motivation and action. In social learning theory (Bandura, 1977b, in press), self-directedness operates through a self system that comprises cognitive structures and subfunctions for perceiving, evaluating, motivating, and regulating behavior. An important, cognitively based source of This research was supported by Public Health Research Grant M-5162 from the National Institute of Mental Health. We are deeply indebted to the many people who assisted us in this project: Ruthe Lundy and Jack Gibbany of the Palo Alto Unified School District arranged the necessary research facilities. School principals Jerry Schmidt, Gene Tankersley, John Tuomy, Roger Wilder, and their staffs offered whatever help was needed to facilitate the research. Jamey Friend and Barbara Searle of the Stanford Institute for Mathematical Studies in the Social Sciences furnished us with invaluable information on mathematical subfunctions, which served as the basis for the self-instructional material. Finally, we owe a debt of appreciation to Debby Dyar and Linda Curyea for their generous and able assistance in the conduct of the experiment. Requests for reprints should be sent to Albert Bandura, Department of Psychology, Building 420, Jordan Hall, Stanford University, Stanford, California 94305.
self-motivation relies on the intervening processes of goal setting and self-evaluative reactions to one's own behavior. This form of self-motivation, which operates largely through internal comparison processes, requires personal standards against which to evaluate ongoing performance. By making self-satisfaction conditional on a certain level of performance, individuals create selfinducements to persist in their efforts until their performances match internal standards. Both the anticipated satisfactions for matching attainments and the dissatisfactions with insufficient ones provide incentives for self-directed actions. Personal goals or standards do not automatically activate the evaluative processes that affect the level and course of one's behavior. Certain properties of goals, such as their specificity and level, help to provide clear standards of adequacy (Latham & Yukl, 1975; Locke, 1968; Steers & Porter, 1974). Hence, explicit goals are more likely than vague intentions to engage self-reactive influences in any given activity. Goal proximity, a third property, is especially critical because the more closely referential standards are related to ongoing behavior, the greater the likelihood that self-influences will be activated during the process. Some 586
PROXIMAL SELF-MOTIVATION
suggestive evidence exists that the impact of goals on behavior is indeed determined by how far into the future they are projected (Bandura & Simon, 1977; Jeffery, 1977). In the social learning view, adopting proximal subgoals for one's own behavior can have at least three major psychological effects. As already alluded to, such goals have motivational effects. One of the propositions tested in the present experiment is that selfmotivation can be best created and sustained by attainable subgoals that lead to larger future ones. Proximal subgoals provide immediate incentives and guides for performance, whereas distal goals are too far removed in time to effectively mobilize effort or to direct what one does in the here and now. Focus on the distant future makes it easy to temporize and to slacken efforts in the present. Proximal subgoals can also serve as an important vehicle in the development of selfpercepts of efficacy. Competence in dealing with one's environment is not a fixed act or simply knowing what to do. Rather, it involves a generative capability in which component skills must be selected and organized into integrated courses of action to manage changing task demands. Operative competence thus requires flexible orchestration of multiple subskills. Self-efficacy is concerned with judgments about how well one can organize and execute courses of action required to deal with prospective situations containing many ambiguous, unpredictable, and often stressful elements. Self-percepts of efficacy can affect people's choice of activities, how much effort they expend, and how long they will persist in the face of difficulties (Bandura, 1977a; Brown & Inouye, 1978; Schunk, 1981). Without standards against which to measure their performances, people have little basis for judging how they are doing or for gauging their capabilities. Subgoal attainments provide indicants of mastery for enhancing self-efficacy. By contrast, distal goals are too far removed in time to provide sufficiently clear markers of progress along the way to ensure a growing sense of selfefficacy. The processes underlying the development of intrinsic interest may similarly be gov-
587
erned, at least in part, by goal proximity. Most of the activities that people enjoy doing for their own sake originally had little or no interest for them. Young children are not innately interested in singing operatic arias, playing tubas, deriving mathematical equations, writing sonnets, or propelling heavy shotput balls through the air. However, through favorable continued involvement, almost any activity can become imbued with consuming significance. One can posit at least two ways in which proximal goals might contribute to enhancement of interest in activities. When people aim for, and master, desired levels of performance, they experience a sense of satisfaction (Locke, Cartledge, & Knerr, 1970). The satisfactions derived from subgoal attainments can build intrinsic interest. When performances are gauged against lofty, distal goals, the large negative disparities between standards and attainments are likely to attenuate the level of self-satisfaction experienced along the way. Conceptual analyses of intrinsic interest within the framework of both self-efficacy theory (Bandura, 1981) and intrinsic motivation theory (Deci, 1975; Lepper & Greene, 1979) assign perceived competence a mediating role. A sense of personal efficacy in mastering challenges is apt to generate greater interest in the activity than is selfperceived inefficacy in producing competent performances. To the extent that proximal subgoals promote and authenticate a sense of causal agency, they can heighten interest through their effects on perception of personal causation. Investigations of intrinsic interest have been concerned almost exclusively with the effects of extrinsic rewards on interest when it is already highly developed. Although results are somewhat variable, the usual finding is that rewards given regardless of quality of performance tend to reduce interest, whereas rewards for performances signifying competence sustain high interest (Boggiano & Ruble, 1979; Enzle & Ross, 1978; Lepper & Greene, 1979; Ross, 1976). The controversy over the effects of extrinsic rewards on preexisting high interest has led to a neglect of the issue of how intrinsic interest is developed when it is lacking. One of the present
588
ALBERT BANDURA AND DALE H. SCHUNK
study's aims was to test the notion that proximal subgoals enlist the type of sustained involvement in activities that build self-efficacy and interest when they are absent. Children who displayed gross deficits in mathematical skills and strong disinterest in such activities engaged in self-directed learning over a series of sessions. They pursued the self-paced learning under conditions involving either proximal subgoals, distal goals, or bids to work actively without any reference to goals. It was predicted that self-motivation through proximal subgoals would prove most effective in cultivating mathematical competencies, self-percepts of efficacy, and intrinsic interest in mathematical activities. For reasons given earlier, distal goals were not expected to exceed bids to work actively in promoting changes. It was hypothesized further that strength of self-efficacy would predict subsequent accuracy on mathematical tasks and level of intrinsic interest. Method
Subjects The subjects were 40 children of predominantly middle-class backgrounds, ranging in age from 7.3 to 10.1 years, with a mean age of 8.4 years. There were 21 males and 19 females distributed equally by age and sex across conditions. Children were drawn from six elementary schools. As an initial screening procedure, teachers identified children in their classes who displayed gross deficits in arithmetic skills and low interest in such activities. The pretreatment procedures were administered to each of the identified children by one of two testers (a male and a female) to determine whether the children's arithmetic skills were sufficiently deficient to qualify for the experiment.
Pretreatment Measures The study was presented to the children as a project aimed at gaining understanding of how arithmetic skills are acquired. To reduce further any evaluative concerns, they were informed that the project was being conducted in several schools and that their work would be treated in full confidence. Mathematical performance test. The performance pretest consisted of 25 subtraction problems graded by level of difficulty. The test problems, which ranged from two to six columns, were specifically designed so as to tap each of seven subtraction operations that were included in the treatment phase of the study.
Of the 25 pretest problems, 8 were similar in form and difficulty to some of the types of items used in the subsequent treatment phase. To test for generalization effects, 17 problems required application of the various subtraction operations to problem forms that were more complex than those children would encounter in the selfinstructional phase. For example, in treatment they learned how to borrow once or twice from zero in, at most, four-column problems, whereas a generalization item would require borrowing from three consecutive zeros in a six-column set. To add a further element of complexity, all the test problems were cast in a form in which the minuend and the difference between the minuend and the subtrahend were provided so that children had to solve the subtrahend. Children were presented the set of 25 subtraction problems one at a time on separate pages and were instructed to turn each page over after they had solved the problem or had chosen not to work at it any longer. The tester recorded the time spent on each problem. The measure of competence in subtraction was the number of problems in which the children applied the correct subtraction operation. Pilot work in the development of the test procedures revealed that children who do not fully comprehend subtraction operations fail to grasp the nature of their deficiency because they faithfully apply an erroneous algorithm. When presented with a subtraction problem, they simply subtract the smaller number from the larger one in each column regardless of whether the smaller number is the minuend or the subtrahend. To address this problem at the outset of the experiment proper, after children completed the arithmetic pretest they compared their solutions to the correct answers. However, children performed the posttest without feedback of accuracy. Since this study centered on motivational processes by which competencies, perceived efficacy, and interest can be developed when they are lacking, children who solved more than four problems were excluded. The selected sample of children indeed exhibited gross deficits; one third could not solve a single problem, and another third could only solve one. The children's substantial quantitative deficiencies were further confirmed by standardized measures of their mathematical ability obtained from the school district on three subtests of the Metropolitan Achievement Test (Durost, Bixler, Wrightstone, Prescott, & Balow, 1970). They occupied the bottom percentile ranks in computation (22), concepts (27), and problem solving (22). Children in the various treatment conditions did not differ in this respect. Self-efficacy judgment. Before measuring perceived mathematical efficacy, children performed a practice task to familiarize them with the efficacy assessment format. The tester stood at varying distances from the children and asked them to judge whether they could jump selected distances and to rate on a 100-point scale the degree of certainty of their perceived capability. In this concrete way, children learned how to use numerical scale values to convey the strength of their self-judged efficacy. In measuring strength of mathematical self-efficacy, children were shown, for 2 sec each, 25 cards containing
PROXIMAL
SELF-MOTIVATION
pairs of subtraction problems of varying difficulty. This brief exposure was sufficient to portray the nature of the tasks but much too short to even attempt any solutions. After each sample exposure, children judged their capability to solve the type of problem depicted and rated privately the strength of their perceived efficacy on a 100-point scale, ranging in 10-unit intervals from high uncertainty through intermediate values of certainty to complete certitude. The higher the scale value, the stronger the perceived self-efficacy. The measure of strength of self-efficacy was obtained by dividing the summed magnitude scores by the total number of problems.
Self-Instructional Material Research conducted at the Stanford Institute of Mathematical Studies has shown that competence in subtraction requires several subskills (Friend & Burton, 1981). These include subtracting a number from a larger one, subtracting zero, subtracting a number from itself, borrowing once, borrowing caused by zero, borrowing twice, borrowing from 1, and borrowing from zero. Seven sets of instructional material were designed, which incorporated the various subtraction operations. The material was organized in such a way that children could work independently at their own pace over a series of sessions. The format of each instructional set was identical. The first page of each set contained a full explanation of the relevant subtraction operation, along with two examples illustrating how the solution strategies are applied. The following six pages contained sets of problems to be solved using the designated operation. Pretesting showed that if children worked at a steady pace, they could complete each self-instructional set in about 25 min.
Procedure for Self-Directed Learning One of three experimenters (one male and two females) brought the children individually, at slightly staggered times, into the study room, where they were seated in different locations, facing away from each other to preclude any visual contact. Both the experimenter and the schools from which the children were drawn were the same across treatment conditions. The entire set of instructional materials was placed face down on the table. The children were informed that they could work on these subtraction problems for seven 30min. sessions. In describing the procedure for self-directed learning, the experimenter turned over the first page, which explained the subtraction operation for the first six-page set. Children were told that whenever they came to a page of instructions, they should bring it to the experimenter who would read it to them. Then they should solve, on their own, the subtraction problems contained on the succeeding pages. If children asked for further assistance with the instructions, the experimenter simply reread the relevant sections of the instructions but never supplemented them in any way. Since the instructions were self-explanatory and the importance of children
589
working on their own was underscored, they rarely sought the experimenter's attention during the sessions. The self-directed study was conducted on consecutive school days. At the end of each session, children marked where they had stopped and simply resumed work at that point in the subsequent session. The situational arrangement for the instructional phase of the study was designed to leave the initiative to the children, thus allowing leeway for self-directedness and self-motivation to exert their effects. By working independently but in a group setting, none of the children was the focus of attention. The seating arrangement and sequential entry precluded communication between the children. After delivering the instructions, the experimenter retired to a table away from the children and remained as unobtrusive as possible throughout the sessions. By having children in different treatments pursue separately the self-directed learning in the same setting at the same time, social and situational factors that might otherwise vary were comparable across treatment conditions.
Treatment Conditions Children were assigned randomly to one of three treatment conditions or to a nontreated control group. The instructions, format, and materials for the self-directed study were identical across treatment conditions except for variations in goal setting. Proximal goals. For children in the proximal-goal treatment, the experimenter suggested that they might consider setting themselves a goal of completing at least six pages of instructional items each session. To give some salience to a continuing goal orientation, the suggestion of proximal goals was made at the beginning of the second session as well. There was no further mention of goals thereafter. Distal goals. For children assigned to the distal-goal treatment, the experimenter suggested that they might consider setting themselves the goal of completing the entire 42 pages of instructional items by the end of the seventh session, which comprised a total of 258 problems. In both treatment conditions, the goals were mentioned suggestively rather than prescriptively so as to leave the goal-setting decision to the children. This mode of goal structuring was used for two reasons. First, it was designed to increase children's self-involvement in the instructional task. Second, choice increases the level of personal responsibility and commitment to goals. No goals. A third group of children pursued the selfdirected learning without any reference to goals. However, they were told to try to complete as many pages of instructional items as possible as they went along. The reasons for including this particular condition were twofold: to provide a control for the effects of self-directed instruction alone and to equate the groups for the social suggestion that they work productively. No treatment. A fourth group of children was administered the full set of assessment procedures without any intervening exposure to the instructional material. This group provided a control for any possible effects of testing and concomitant classroom instruction.
590
ALBERT BANDURA AND DALE H. SCHUNK
Posttreatment Assessment The procedures used in the pretreatment phase of the study were readministered on the day following completion of the fourth session. This intermediate point was selected to gauge the effects of goal proximity on the development of skill, self-efficacy, and intrinsic interest within an identical length of time. Had children been tested after completing the entire program of study, the posttreatment changes would have been confounded by variations in the amount of time different children required to complete the self-instruction. Children's mathematical self-efficacy was measured at the end of treatment and after the posttest of subtraction performance. The self-efficacy scores obtained at the end of treatment were used to gauge the value of self-efficacy judgment in predicting subsequent arithmetic performance. Since posttest performance can affect perceived efficacy, the measure of self-efficacy obtained following the arithmetic posttest was related to the subsequent measure of intrinsic interest. Each of the 25 pairs of efficacy assessment items, which were the same as those used in the pretreatment assessment, corresponded in form and difficulty level to a subtraction problem in the performance test but involved different sets of numbers. As noted previously, most of the test problems were more complex than the ones included in the treatment phase of the study. Because the test of self-efficacy tapped new applications of cognitive operations, children had to rely on generalizable perceptions of their mathematical capabilities in making their efficacy judgments. A parallel form of the performance test used in the pretest was devised for the posttreatment assessment of mathematical competence. This eliminated any possible effects due to familiarity with problems. Both forms were administered in a counterbalanced order to a sample of 17 children who were not participants in the formal study. The alternate forms yielded highly comparable scores (r = .87). Children's intrinsic interest in subtraction problems was measured in a separate session scheduled the day after the posttreatment assessment. The tester explained that she/he had another task the children could do. Their attention was then drawn to two stacks of 10 pages each. One stack contained 60 subtraction problems of varying levels of difficulty; the other stack contained rows of digit-symbol problems adapted from the Wechsler Intelligence Scale for Children (Wechsler, 1974). The latter task involved filling in rows of empty squares with symbols corresponding to the digits appearing above each square. The tester stressed that the children should feel free to decide whether they wanted to work on one, or the other, or both tasks. It was further emphasized that it was up to them to decide how much time they wanted to spend on each activity. The children worked alone until 25 min. had elapsed. The number of subtraction problems the children solved under these permissive choice conditions constituted the measure of intrinsic interest. All of the assessment procedures were administered individually by the same tester in both phases of the study. To control for any possible bias, the testers had
no knowledge of the conditions to which the children had been assigned. After the experiment was concluded, all children, including the controls, pursued self-directed instruction to completion under proximal subgoals to provide maximal benefits for all participants.
Results No significant sex differences were found on any of the measures at either the pretest or posttest assessments, since the sample was confined to children with gross arithmetic deficits. In the posttest assessment, children showed comparable gains in self-efficacy and arithmetic performance on generalization problems and on the types of items used in the treatment phase. Having mastered particular subtractive operations on simpler exemplars, children applied them accurately to more complex forms. The data were therefore pooled across sex and class of item for the primary analyses. Analyses of variance were computed on the different sets of data, with phases of the experiment and treatment conditions representing the main factors. At the pretest phase, the groups did not differ on any of the measures. Significant intergroup differences obtained in the posttreatment phase were analyzed further, using the NewmanKeuls multiple-comparison method. Table 1 shows the significance levels of the treatment effects, the changes achieved by children within each condition, and comparisons between treatment conditions. Perceived
Self-Efficacy
The strength of children's perceived mathematical efficacy at different phases of the experiment is presented graphically in Figure 1. Analysis of these data shows the main effect of treatment, F(3, 36) = 10.13, p < .001, and the interaction between treatment and experimental phases, F(6, 72) = 5.96, p < .001, to be highly significant. Intragroup comparisons of changes in strength of self-efficacy, evaluated by the t test for correlated means, yielded no significant differences for children in the control group (Table 1). Those who had the benefit of proximal subgoals substantially increased
591
PROXIMAL SELF-MOTIVATION
Table 1 Significance of Intergroup Differences and Intragroup Changes
Measure
Proximal vs. distal
Proximal vs. no goals
Proximal
vs. control
Distal vs. no goals
Distal control
vs.
No goals vs. control
Intergroup comparisons (Newman-Keuls comparisons) Strength of Selfefficacy Post, Post2
<.05 <.01
<.05 <.05
<.01 <.01
ns ns
<.05 <.05
<.01
Arithmetic performance
<.01
<.01
<.01
IM
<.01
<.01
IM
IM
IM
ns
ns
ns
IM
Persistence Easy problems Difficult problems
ns
ns
<.05
IM
<.05
<.Q5
Intrinsic interest
<.05
<,05
<.05
ns
ns
ns
Accuracy of SelfAppraisal of Efficacy
<.05
<.05
<.05
iw
ns
ns
Intragroup changes 0 tests)
No Proximal
Distal
Control
Strength of Self-efficacy Pre vs. Post, Pre vs. Post2 Posti vs. Post2
4.69*** 5.90**** 3.55***
2.93** 2.15* 2.75**
2.16* 3.70*** 1.18
0.01 0.78 1.12
Arithmetic Performance
12.62****
3.17***
4.27***
1.01
0.26 1.41
0.32 1.76
4.12*** 3.34***
Persistence Easy problems Difficult problems
* p< .10. ** p < .05. *** p < .01.
0.72 4.57*** "p< .001.
their perceived self-efficacy and exhibited even further gains following the performance posttest. Children oriented toward distal goals displayed a moderate increase in selfefficacy but a small decline after the posttest. Self-directed learning without goals produced a modest increase at a borderline level of significance. In separate comparisons between treatments, the proximal group exceeded all others in strength of perceived self-efficacy, as measured both before and after the behav-
ioral posttest (Table 1). Children in the distal condition also exceeded the controls in self-efficacy, but they did not differ significantly from those who set no goals for themselves. Children in the latter condition judged their mathematical efficacy more highly than did the controls after but not before the performance posttest. Mathematical Performance Figure 1 presents the mean scores obtained on the subtraction performance test
592
ALBERT BANDURA AND DALE H. SCHUNK 90
•—• PROXIMAL GOALS •--• DISTAL GOALS o-o NO GOALS 80 h o—o CONTROL
O 70 u. Ul
IL 60 50
30
1 PRETEST
2
PRETEST
POSTTEST
POSTTEST
Figure 1. The left panel shows the strength of children's self-percepts of arithmetic efficacy at the beginning of the study (pretest), and before (Posti) and after (Postz) they took the subtraction posttest. The right panel displays the children's level of achievement on the subtraction tests before and after the self-directed learning.
by children in the various conditions. The main effect of treatment was highly significant, F(3, 36) = 12.80, p < .001, as was the interaction between treatment and experimental phases, F(3, 36) = 12.55, p < .001. Self-directed instruction promoted mastery of subtractive operations in all three groups, whereas the controls remained at a loss on how to subtract numbers from each other (Table 1). In pairwise comparisons, children who had employed proximal subgoals surpassed all the other groups in subtractive skills. Children who engaged in self-directed learning either with distal or no goals did not differ significantly from each other, but both groups outperformed the controls. In the above measure, children received partial credit if they applied the appropriate subtractive operations but made a minor error in deriving or in recording the answer. The identical pattern of results is obtained on scores using a stringent criterion requiring perfect accuracy on all counts. In comparing the children's subtractive skills before and after treatment, all three groups that engaged in self-directed instruction achieved significant gains beyond the p < .001 level of significance; whereas the controls, who
solved only 5% of the problems in pretest and only 8% in posttest, remained grossly deficient in this regard. Contrasts between the means of the different treatment conditions show the children in the proximal condition to be much more skilled than those in the distal (p < .01), no-goals (p< .01), or control (p < .01), conditions. Children who pursued the selflearning with distal (/?<.01) or no goals (p < .01) were also more skilled than the controls, but the former two groups did not differ from each other. Persistence Children who gain high self-efficacy through skill acquisition solve problems readily and, therefore, need not spend much time on them. An aggregate measure of persistence spanning the entire range of difficulty is not too meaningful because long perseverence times on hard problems are offset by rapid solutions of less difficult ones. Changes in persistence were, therefore, analyzed separately for problems at two levels of difficulty: The difficult set of problems required two or more borrowing operations, whereas the easier set involved either no bor-
593
PROXIMAL SELF-MOTIVATION
rowing or, at most, only one smaller minuend. Analysis of persistence on the easy arithmetic items revealed no differences except for the controls, who were significantly less persevering (-31%) when tested again. However, in the efforts expended on difficult arithmetic items, the proximal children were markedly more persistent after treatment than before (+90%), the distal (+22%) and the no-goals children (+39%) were moderately more persistent, whereas the controls slackened their efforts (-27%) in the posttest. This differential pattern of perseverence yielded a highly significant Treatment X Phases interaction, F(3, 36) = 5.67, p < .005. Analyses of intragroup changes presented in Table 1 show that the increased perseverence of the proximally self-motivated children and the diminished effort of the controls are highly significant. Multiple comparisons among groups in the posttest assessment disclose that children in each of the treatment conditions, although not differing from each other, were all significantly more persevering than were the controls (Table 1). Intrinsic Interest The role of goal proximity in the development of intrinsic interest may be seen in Figure 2. Analysis of variance of the number of subtraction problems that children chose to solve on their own yielded a significant treatment effect, F(3, 36) = 3.57, p < .05. Inspection of Table 1 shows that children in the proximal subgoal condition exceeded all three comparison groups, which did not differ from each other. Indeed, 90% of the children who developed their arithmetic skill through the aid of proximal subgoals performed subtraction problems under the freechoice conditions; whereas only about 40% of the children in the other groups did so. The involvement in arithmetic problems displayed by the proximal children was not at the detriment of the competing activity. Children in all groups performed a comparable number of digit-symbol items and did not differ in this respect (F = 0.77). It would seem that experience with proximal self-motivators enhances the total level of subse-
14
12
£ 10 UJ
t-
z
O 8
z cc
PROXIMAL GOALS
DISTAL GOALS
NO GOALS
CONTROL
Figure 2. Average number of subtraction problems children in the different conditions chose to solve when given free choice of activities.
quent productivity, since children in this condition were as prolific on the competing task and more so on the arithmetic one. Progress in Self-Directed Learning The average length of time it took children to complete each lesson was 21, 29, and 30 min. for the proximal, distal, and no-goals conditions, respectively. Thus, proximal selfmotivators produced more rapid mastery of the subject matter than did distal ones, F(l, 27) = 3.94, p < .10, or self-instruction without goals, F(l, 27) = 5.44, p < .05. By the end of the four sessions, the percentage of the total instructional material completed was 74% in the proximal condition, 55% in the distal condition, and 53% in the condition involving no goals. Although proximal subgoals, compared to distal goals, F(l, 27) = 3.66, p < .10, and no goals, F(l, 27) = 4.67, p < .05, fostered greater mastery of subtractive operations, distal goals had no significant effect either on rate or level of self-directed instruction. Congruence Between Self-Efficacy Judgment and Performance Congruence indices can be computed by comparing efficacy judgments at the end of
594
ALBERT BANDURA AND DALE H. SCHUNK
treatment with subsequent posttest performance of subtraction problems of comparable form and difficulty. In this procedure, judgments of self-efficacy are dichotomized into positive and negative instances, based on a selected cutoff strength value. Instances of congruence occur when children judged themselves capable of solving a given level of problem and, in fact, solved it or judged themselves incapable and then failed the same class of problem. Mismatches between efficacy judgments and performances (i.e., judged efficacy for failed items and judged inefficacy for solved items) represent instances of incongruence. Congruence indices were calculated separately for efficacy cutoff values at different levels of strength. When a low-strength value is selected as the criterion of self-judged efficacy, a weak sense of efficacy (e.g., 20) is treated like complete certitude (100). Such a low criterion could produce artifactual mismatches. If the criterion were set near the maximal strength value (80), reasonably high levels of self-judged efficacy (e.g., 70) would be defined as inefficacy. This too would produce artifactual discrepancies. For the mathematical performances examined in this study, an efficacy strength of 40, which reflects a moderate degree of assurance, provided the optimal cutoff criterion. Results of the congruence analysis disclose that the conditions of treatment affected the level of accuracy with which children appraised their mathematical efficacy, F(3, 36) = 3.06, p < .05. Children in the distal (54%), no-goals (51%), and control (60%) conditions displayed moderate congruence between their self-judged efficacy and their performance. In contrast, children who developed their skills under proximal subgoals were highly accurate in their selfappraisals of efficacy (80%). Table 1 shows that in accurateness of self-appraisal, the proximally self-motivated children exceed other experimental groups, which do not differ significantly from each other. In most instances the children erred by overestimating their capabilities. Correlational Analyses Correlational analyses were carried out to provide additional information on the rela-
tionship between theoretically relevant variables. Product-moment correlations were calculated separately within groups, and when they did not differ significantly, they were averaged by means of an r to z transformation. That skill acquisition builds self-efficacy receives support in the data. The more selfinstructional material the children mastered, the stronger was their sense of mathematical self-efficacy, r(28) = .42, /j<.01. Performances that are readily achieved suggest a higher level of self-ability than do analogous attainments gained through slow, heavy labor. Consistent with this expectation, the faster the children completed each lesson, the more efficacious they judged themselves to be, r(28) = .32, /><.05. Standardized measures of mathematical competence, based on the Metropolitan Achievement Test, did not predict rate of skill acquisition or degree of self-efficacy enhancement. Both instructional performance, r(28) = .33, p < .05, and strength of self-efficacy, K38) = .42,/><.01, were moderately related to the children's facility in using subtractive operations. However, strength of self-efficacy was a significant predictor of perfect arithmetic accuracy for the total sample, r(38) = .49, /x.OOl, and for the selfinstructed groups, r(28) = .40, p < .025, whereas past self-instructional performance was only marginally related to faultless posttest performance, r(28) = .25, p < .10. Skill acquisition speeds problem solving. This factor attentuated the relationship between self-efficacy and persistence for the treated children who found most of the problems readily soluble and, hence, had no need to spend much time on them. The influence of perceived self-efficacy on perseverence is, of course, best revealed on problems that cannot be solved however hard one tries. Children who doubt their capabilities quit sooner than those who believe they can eventually master the task should they persevere. This condition obtains for the control children whose marked arithmetic deficiencies rendered most of the problems insoluble. For this group, the stronger the children's selfperceived efficacy, the longer they persevered, K8) = .63, p = .025. For all children, high perseverence was accompanied by high
PROXIMAL SELF-MOTIVATION
performance attainments on the more difficult problems, r(38) = .51, p < .001. Even the easy problems were exceedingly difficult for the control children, and here too, persistence was related to performance success, K8) = .61, p < .001. The relationship of perceived self-efficacy to intrinsic interest can be analyzed in several ways. It may require at least moderately high self-efficacy to generate and sustain interest in an activity, but interest is not much affected by small variations above or below the threshold level. To test this threshold notion, interest scores were correlated with number of self-percepts of efficacy that matched or exceeded an efficacy strength of 40, which was previously shown to be the optimal criterion in the congruence analysis. The results disclose that the higher the level of self-efficacy at the end of the posttest, the greater the interest shown in arithmetic activities, r(38) = .27, p < .05. An alternative possibility is that intrinsic interest is linearly related to strength of selfefficacy. Correlation of the latter measures shows that variations in mean strength of self-efficacy covaried with interest in the control and the no-goals conditions, K18) = .39, p < .05, but not in the goal-setting treatments. Although interest was positively related to self-percepts of efficacy derived from performance attainments in treatment, it was uncorrelated with the performance attainments themselves. However, the standardized measures of competence in mathematical subfunctions emerged as significant correlates. The more competent children were at mathematical computation, H38) = .42, p < .01, and problem solving, K38) = .58, p < .001, the more subtraction problems they completed in the free-choice situation. Discussion Results of the present study confirm the influential role of proximal self-motivators in the cultivation of competence, self-percepts of efficacy, and intrinsic interest. Children who set themselves attainable subgoals progressed rapidly in self-directed learning, achieved substantial mastery of mathemat-
595
ical operations, and heightened their perceived self-efficacy and interest in activities that initially held little attraction for them. Efforts to clarify how goal proximity operates in self-regulatory mechanisms ordinarily present difficulties because even though encouraged to set themselves distal goals, people are prone to convert them into more aidful proximal ones. They simply fractionate desired future accomplishments into attainable daily subgoals (Bandura & Simon, 1977). In the present experiment, children could not transform distal into proximal self-motivators because not knowing how to divide, they could not partition the entire instructional enterprise into equivalent subunits. Results of the combined studies attest to the motivating potential of proximal goals, whether they are suggested or spontaneously generated, or whether the self-sustained behavior is tractable or very difficult to produce. Judgment of mathematical self-efficacy by children just beginning to understand the requisite cognitive skills is no simple matter. With very brief exposure to sample items, it is hard to discriminate among different levels of task difficulty. This is because the complexity of some of the subtractive operations is not instantly apparent from what is most readily observable. When complex operations are imbedded in seemingly easy problems, which is often the case, appearances can be quite misleading. In such situations, incongruities between perceived selfefficacy and action may stem from misperceptions of task demands as well as from faulty self-knowledge. Moreover, solving problems typically requires applying multiple operations. Even if they were readily recognizable, judgment of personal capabilities for a given type of task is complicated if some of the constituent operations are thoroughly mastered and others are only partially understood. Selective attention to mastered elements highlights competencies; whereas focus on what is less well understood highlights shortcomings. Even equal attentiveness to all aspects of the task will produce some variance in judgments of self-efficacy, depending on how much weight is given to the differentially mastered operations. Given these complexities, it is not sur-
596
ALBERT BANDURA AND DALE H. SCHUNK
prising that children sometimes overestimated their capabilities, especially on tasks that appeared deceptively simple. However, it is noteworthy that in addition to its other benefits, goal proximity fosters veridical selfknowledge of capabilities. Children who guided and judged their progress in terms of proximal subgoals were highly accurate in their self-appraisals. In contrast, skill development under conditions in which progress toward competence is somewhat ambiguous does not improve self-knowledge. Despite the fact that children in the distal and no-goal conditions acquired new information about mathematics and applied it repeatedly, the accurateness of their self-appraisals was no better than that of the controls, who did not have the benefit of substantial performance information for constructing their self-knowledge. The above results are consistent with previous findings that judgments of self-efficacy are not simply reflectors of past performance (Bandura, 1977a, 1977b). Rather they reflect an inferential process in which the selfability inferences drawn from one's performances vary, depending on how much weight is placed on personal and situational factors that can affect how well one performs. The evaluative standards against which ongoing performances are appraised constitute an additional factor that determines how well people judge their capabilities. Results of microanalyses of congruence between self-efficacy judgment and performance indicate that the optimal cutoff value of efficacy strength varies across activities, depending on the complexity and the variety of skills they- require. Performances that draw on only a few skills reduce the likelihood of overestimating personal capabilities by overweighting a mastered subpart. Hence, lower efficacy strength values can be predictive of success (Bandura, 1977a). In activities that depend on diverse subskills, knowledge of some of them, especially the more directly observable ones, raises the level of assurance that one might be able to perform successfully. Consequently, somewhat higher cutoff values of efficacy strength become predictive of success. Research on intrinsic interest has centered
primarily on how extrinsic incentives affect high interest when it is already present rather than on how to develop it when it is lacking. It is the latter problem that presents major challenges, especially when avoidance of activities essential for self-development reflects antipathy arising from repeated failure rather than mere disinterest. The present findings lend support to the general thesis that skills cultivated through proximal standards of competency build interest in disvalued activities. When progress is gauged against distal goals, similar accomplishments may prove disappointing because of wide disparities between current performance and lofty future standards. Consequently, interest fails to develop, even though skills are being acquired in the process. Perceived self-efficacy was accompanied by high-performance attainments and perseverence under conditions in which such a relationship would be expected to obtain. Regardless of conditions of treatment, persistency increased the likelihood of success. There was some evidence to indicate that faultless arithmetic performance was better predicted from self-efficacy than from behavioral attainments in treatment. However, caution should be exercised in judging causal contributions from correlations because some of the measures reflect continuously interactive processes rather than discrete sequential ones. Consider, for example, the question of directionality of influence in obtained relationships between treatment performance, posttreatment self-efficacy, and posttest performance. Self-efficacy judgments are unconfounded by future posttest performance, but it is highly unlikely that self-percepts of efficacy played no role whatsoever in performance attainments during the self-directed learning. Judgments of one's capabilities can affect rate of skill acquisition, and performance mastery, in turn, can boost self-efficacy in a mutually enhancing process. It is not as though self-efficacy affects future performances in the posttest but does not affect earlier performances in the treatment phase. The causal contribution of perceived selfefficacy to performance is most clearly revealed in studies in which self-percepts of
PROXIMAL SELF-MOTIVATION
efficacy are developed solely through vicarious or cognitive means that entail no overt performance (Bandura & Adams, 1977; Bandura, Adams, & Beyer, 1977; Bandura, Adams, Hardy, & Howells, 1980; Bandura, Reese, & Adams, Note 1). Perceived selfefficacy, instated symbolically, predicts well the pattern of performance successes and failures on specific tasks. A further issue addressed in this research concerns the relationship between perceived self-efficacy and intrinsic interest. It was mainly children in the proximally self-motivated condition, all of whom felt highly efficacious, who displayed the notable level of intrinsic interest. In contrast, children in the other conditions generally expressed selfdoubts concerning their capabilities and showed little spontaneous interest in solving arithmetic problems. Regardless of treatment conditions, self-percepts of moderate to high strength were positively related to interest. Intrinsic interest seems to covary most closely with the more long-standing indicants of actual or perceived competence, that is, the standardized measures of competence in mathematical subfunctions and perceived self-efficacy in groups whose preexisting self-percepts were either unaltered or changed only marginally. These findings raise the interesting possibility that some temporal lag exists between newly acquired self-efficacy and corresponding growth of interest. The link between boosts in perceived self-efficacy and sustained involvement in challenging activities is now well established across a wide range of behavioral domains (Bandura, 1981; Brown & Inouye, 1978; Weinberg, Gould, & Jackson, 1979; Schunk, 1981; Condiotte & Lichtenstein, in press). But it may require mastery experiences over a period of time before the selfefficacy derived from progressive successes creates strong interest in activities that were disvalued or even disliked. If, in fact, effects follow such a temporal course, then increased interest would emerge as a later rather than as an instant consequent of enhanced self-efficacy. Because of the theoretical import of the link between self-efficacy and interest, both the threshold and the tem-
597
poral lag effects warrant systematic investigation. Reference Note 1. Bandura, A., Reese, L., & Adams, N. E. Microanalysis of action and fear arousal as a function of differential levels of perceived self-efficacy. Unpublished manuscript, Stanford University, 1981.
References Bandura, A. Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 1977, 84, 191-215. (a). Bandura, A. Social learning theory. Englewood Cliffs, N.J.: Prentice-Hall, 1977. (b) Bandura, A. Self-referent thought: A developmental analysis of self-efficacy. In J. H. Flavell & L. Ross (Eds.), Social cognitive development: Frontiers and possible futures. Cambridge, England: Cambridge University Press, 1981. Bandura, A. The self and mechanisms of agency. In J. Suls (Ed.), Psychological perspectives on the self (Vol. 1). Hillsdale, N.J.: Erlbaum, in press. Bandura, A., & Adams, N. E. Analysis of self-efficacy theory of behavioral change. Cognitive Therapy and Research, 1977, /, 287-308. Bandura, A., Adams, N. E., & Beyer, J. Cognitive processes mediating behavioral change. Journal of Personality and Social Psychology, 1977, 55, 129-139. Bandura, A., Adams, N. E., Hardy, A. B., & Howells, O. N. Tests of the generality of self-efficacy theory. Cognitive Therapy and Research, 1980, 4, 39-66. Bandura, A., & Simon, K. M. The role of proximal intentions in self-regulation of refractory behavior. Cognitive Therapy and Research, 1977, 1, 177-193. Boggiano, A. K., & Ruble, D. N. Competence and the overjustification effect: A developmental study. Journal of Personality and Social Psychology, 1979, 37, 1462-1468. Brown, I., Jr., & Inouye, D. K. Learned helplessness through modeling: The role of perceived similarity in competence. Journal of Personality and Social Psychology, 1978, 36, 900-908. Condiotte, M. M., & Lichtenstein, E. Self-efficacy and relapse in smoking cessation programs. Journal of Consulting and Clinical Psychology, (in press). Deci, E. L. Intrinsic motivation. New York: Plenum, 1975. Durost, W. N., Bixler, H. H., Wrightstone, J. W., Prescott, G. A., & Balow, I. H. Metropolitan Achievement Tests: Elementary form F. New York: Harcourt Brace Jovanovich, 1970. Enzle, M. E., & Ross, J. M. Increasing and decreasing intrinsic interest with contingent rewards: A test of cognitive evaluation theory. Journal of Experimental Social Psychology, 1978, 14, 588-597. Friend, J., & Burton, R. Teacher's guide: Diagnostic testing in arithmetic: Subtraction. Palo Alto: Xerox Palo Alto Research Center, 1981. Jeffery, K. M. The effects of goal-setting on self-motivated persistence. Unpublished doctoral dissertation, Stanford University, 1977.
598
ALBERT BANDURA AND DALE H. SCHUNK
Latham, G. P., & Yukl, G. A. A review of research on the application of goal setting in organizations. Academy of Management Journal, 1975, 18, 824-845. Lepper, M. R., & Greene, D. Overjustification research and beyond: Toward a means-ends analysis of intrinsic and extrinsic motivation. In D. Greene & M. R. Lepper (Eds.), The hidden costs of reward. Hillsdale, N.J.: Erlbaum, 1979. Locke, E. A. Toward a theory of task motivation and incentives. Organizational Behavior and Human Performance, 1968, 3, 157-189. Locke, E. A., Cartledge, N., & Knerr, C. S. Studies of the relationship between satisfaction, goal setting, and performance. Organizational Behavior and Human Performance, 1970, 5, 135-158. Ross, M. The self perception of intrinsic motivation. In J. H. Harvey, W. J. Ickes, & R. F. Kidd (Eds.), New
directions in attribution research. Hillsdale, N.J.: Erlbaum, 1976. Schunk, D. Modeling and attributional effects on children's achievement: A self-efficacy analysis. Journal of Educational Psychology, 1981, 73, 93-105. Steers, R. M., & Porter, L. W. The role of task-goal attributes in employee performance. Psychological Bulletin, 1974, 81, 434-452. Wechsler, D. Wechsler intelligence scale for children: Form R. New York: The Psychological Corporation, 1974. Weinberg, R., Gould, D., & Jackson, A. Expectations and performance: An empirical test of Bandura's selfefficacy theory. Journal of Sport Psychology, 1979, /, 320-331. Received June 16, 1980 •